How Pinterest Reduced Costs and Improved Data Consistency with a NewSQL Database | PingCAP
Pinterest is an American visual discovery engine for finding ideas like recipes, home and style inspiration, and more. As of March 2021, it had more than 478 million global monthly active users.
In recent years, we used the HBase ecosystem to serve our critical online applications. But this infrastructure was complex. It had limited functionalities and brought high maintenance and infrastructure costs. We looked for a new storage solution that supports distributed transactions and SQL compatibility.
After we compared YugaByteDB, CockroachDB, and TiDB, an open-source scale-out database, we chose TiDB because it best met our requirements for stability and performance. TiDB helps us reduce system complexity and infrastructure cost and achieve stronger consistency.
In 2013, we introduced HBase into Pinterest. Today, we have about 50 clusters, hosting 10+ PB of data, aggregating 100 million+ queries per second (QPS), and serving 100+ critical use cases. Since it was introduced, HBase has been a building block of our Pinterest infrastructure. Today, it still serves our critical online path.
